New features

OpenStack Yoga

Added full support for OpenStack Yoga with Open vSwitch and Tungsten Fabric 2011 networking back ends.

Starting from 22.5, MOSK deploys all new clouds using OpenStack Yoga by default. To upgrade an existing cloud from OpenStack Victoria to Yoga, follow the Upgrade OpenStack procedure.

For the OpenStack support cycle in MOSK, refer to OpenStack support cycle.

Highlights from upstream supported by Mirantis OpenStack deployed on Yoga

  • [Cinder] Removed the deprecated Block Storage API version 2.0. Instead, use the Block Storage API version 3.0 that is fully compatible with the previous version.

  • [Cinder] Removed the requirement for the request URLs to contain a project ID in the Block Storage API making it more consistent with other OpenStack APIs. For backward compatibility, legacy URLs containing a project ID continue to be recognized.

  • [Designate] Added support for the CERT resource record type enabling new use cases such as secure email and publication of certificate revocation list through DNS.

  • [Horizon] Added support for the Network QoS Policy creation.

  • [Glance] Implemented /v2/images/<image-id>/tasks to get tasks associated with an image.

  • [Ironic] Changed the default deployment boot mode from legacy BIOS to UEFI.

  • [Masakari] Added support for disabling and enabling failover segments. Now, cloud operators can put whole segments into the maintenance mode.

  • [Neutron] Implemented the address-groups resource that can be used to add groups of IP addresses to security group rules.

  • [Nova] Added support for the API microversion 2.90. It enables the users to configure the host name exposed through the Nova metadata service during instances creating or rebuilding.

  • [Octavia] Increased the performance and scalability of load balancers that use the amphora provider when using amphora images built with version 2.x of the HAProxy load balancing engine.

  • [Octavia] Improved the observability of load balancers by adding the PROMETHEUS listeners that expose a Prometheus exporter endpoint. The Octavia amphora provider exposes over 150 unique metrics.

Exposable OpenStack notifications

Implemented the capability to securely expose part of a MOSK cluster message bus (RabbitMQ) to the outside world. This enables external consumers to subscribe to notification messages emitted by the cluster services and can be helpful in several use cases:

  • Analysis of notification history for retrospective security audit

  • Real-time aggregation of notification messages to collect statistics of cloud resource consumption for capacity planning or charge-back

The external notification endpoint can be easily enabled and configured through the OpenStackDeployment custom resource.

Shared Filesystems as a Service

TechPreview

Added MOSK support for the Shared Filesystems service (OpenStack Manila), which enables cloud users to create and manage virtual file shares, so that applications can store their data using common network file sharing protocols, such as CIFS, NFS, and so on.

L3 networking for MOSK control plane

TechPreview

Implemented the ability to enable the BGP load-balancing mode for MOSK underlying Kubernetes to allow distribution of services providing OpenStack APIs across multiple independent racks that have no L2 segments in common.

MKE minor version update to 3.5.5

Based MOSK 22.5 on the Cluster release 12.5.0 that supports Mirantis Kubernetes Engine (MKE) 3.5.5.

Automated configuration of public FQDN for the Object Storage endpoint

The fully qualified domain name (FQDN) for the Object Storage service (Ceph Object gateway) public endpoint is now configurable through just a single parameter in the KaaSCephCluster custom resource, which is spec.cephClusterSpec.ingress.publicDomain. Previously, you had to perform a set of manual steps to define a custom name. If the parameter is not set, the FQDN settings from the OpenStackDeployment custom resource apply by default.

The new parameter simplifies configuration of Transport Layer Security of user-facing endpoints of the Object Storage service.

Enhancements for etcd monitoring

Implemented the following enhancements for etcd monitoring:

  • Introduced etcd monitoring for OpenStack by implementing the Etcd Grafana dashboard and by adding OpenStack to the set of existing alerts for etcd that were used for MKE clusters only in previous releases.

  • Improved etcd monitoring for MKE on MOSK clusters by implementing the Etcd dashboard and etcdDbSizeCritical and etcdDbSizeMajor alerts that inform about the size of the etcd database.

Setting of a custom value for a node label using web UI

Implemented the ability to set a custom value for a predefined node label using the Container Cloud web UI. The list of available node labels is obtained from allowedNodeLabels of your current Cluster release.

If the value field is not defined in allowedNodeLabels, select the check box of the required label and define an appropriate custom value for this label to be set to the node.

Documentation enhancements